A tale of two busses
Even a few years ago, there was a plethora of busses that were used for vision-systems design. VME and multibus boards dominated high-end systems, while the lowly EISA bus was relegated to PC-based systems, where performance was not a determining factor. As you will see from this issue of Vision Systems Design, that`s all changing. The emergence of the PCI bus, in particular, has led to a number of high-performance add-in boards. Indeed, more than 20 companies now manufacture frame grabbers for the PCI bus.
In the Product Focus (p. 40), Contributing Editor Rick Nelson looks at some of the latest PCI frame grabbers and discusses the price/performance trade-offs systems integrators need to consider.
To see how systems developers are using such hardware products in these systems, a number of PC-systems-integration articles are featured in this issue. The first, by Alan Richards, vice president of engineering at Vexcel Imaging (Boulder, CO), explains how to design a scanner capable of imaging rolls of film or negatives at very high resolution (p. 24).
In photoprocessing, too, PC systems are hard at work. On p. 32, Dennis Flanagan, imaging products manager of Optimas (Bothell, WA), describes the design of an OCR system that allows the ID numbers to be read from a roll of film to speed portrait processing.
Of course, PCI-based systems have their limitations. In medical and military systems, large amounts of data need to be displayed and processed. To accommodate these requirements in the medical community, Duke University built an asynchronous-transfer-mode network that supports high-speed data and imaging applications and offers connectivity to existing Ethernet networks (p. 28).
For military applications, where data are collected at high speeds, the VMEbus is the bus of choice. On p. 18, Rick Nelson describes a military sonar processing and display system. To obtain the processing power needed, the sonar device exploits a multiprocessor architecture to provide scalable, high-speed, real-time signal-processing performance.
Finally, in our regular Spotlight on Advanced Technology feature (p. 36), Dave Wilson examines how neural-network-based systems are being built for face-recognition systems. Although these systems are under development, they have yet to discriminate between identical twins.
Andy Wilson Editor